methurlant.safetensors

144 MB · SafeTensor

Requires membership

<metal-hurlant>

Métal Hurlant (literal translation: "Howling Metal," "Screaming Metal") is a French comics anthology of science fiction and horror comics stories. Originally created in 1974, the anthologies ceased publication in 1987 but revived between 2002 and 2004 in multilingual editions, and then again in 2020. How to use Lora weight Based on my testing, I recommend trying all weights between 1 and 2 to see which one works best for your use case. How to Craft Effective Prompts for This Model First of all, I recommend that you take inspiration from the images posted by the author of the model. It is not a plug and play LoRA and requires correct prompting. If your image doesn't include any content/style relevant to the LoRA, then it will have NO effect at all. Please note that the words "image" and "illustration" have HUGE weights and should easily trigger the style in case of trouble. If you're looking to get the best results from this model, here are some tips based on an analysis of the language used in the training data: 1. Focus on Key Subjects and Descriptions: - What to Include: The model is most familiar with terms like "scene", "figure", "creature", "background", and "sky". Descriptive words like "large", "red", "intricate", and "mechanical" also play a big role. - Example Prompt: "Create a detailed scene with a large, red mechanical creature in the foreground." 2. Use Common Phrases: - Why It Helps: The model understands certain word pairs and phrases very well because they appeared frequently in the training captions. Using these can make your prompts more natural and effective. - Example Prompt: "The image depicts a fantastical scene set against a vibrant sky." 3. Incorporate Specific Names or Entities: - Targeted Outputs: If your prompt references specific characters, locations, or objects, the model can generate more focused and relevant results. - Example Prompt: "Illustrate a scene featuring a creature in a mechanical environment." 4. Structure Your Prompts Like Descriptions: - Match the Training Style: The model is used to prompts that describe what’s happening in the image, like “The image depicts...”. Mimicking this style can help the model understand and respond better. - Example Prompt: "The image shows a large creature in an intricate, blue-colored scene." 5. Experiment and Refine: - Iterate on Your Prompts: Start with these suggestions, then tweak the words and phrases to see what works best. Small changes can sometimes lead to big improvements in the results. Example prompts you can try with this model: 1. Prompt for a Fantasy Scene: - "Create a vibrant fantasy scene with a large, red creature standing in the foreground against a mechanical background." 2. Prompt for a Detailed Illustration: - "Illustrate a complex figure set against an intricate, blue sky with various mechanical structures in the background." 3. Prompt Featuring a Specific Character: - "Depict the character Moebius exploring a surreal, otherworldly landscape filled with intricate details and a vibrant color palette." 4. Prompt for a Mechanical Creature: - "Show a large, mechanical creature dominating the scene, with intricate patterns and a deep red color palette." 5. Prompt for a Dreamlike Environment: - "Design a dreamlike scene with floating orbs and clouds in the sky, featuring a detailed figure surrounded by an ethereal atmosphere." Dataset Information Top 20 nouns and adjectives by frequency in the dataset: - image: 147 - scene: 80 - figure: 59 - creature: 58 - large: 55 - red: 54 - - sky: 50 - various: 45 - foreground: 44 - intricate: 43 - - palette: 42 - figures: 41 - blue: 40 - overall: 36 - structure: 34 - mechanical: 33 - illustration: 30 - face: 30 Top 20 collocations by frequency in the dataset: - ('to', 'be') - ('The', 'image') - ('appears', 'to') - ('image', 'depicts') - ('color', 'palette') - ('there', 'are') - ('In', 'the') - ('depicts', 'a') - ('the', 'foreground') - ('theres', 'a') - ('dominated', 'by') - ('The', 'overall') - ('seems', 'to') - ('palette', 'is') - ('a', 'large') - ('There', 'are') - ('graphic', 'novel') - ('shades', 'of') - ('is', 'dominated') - ('scene', 'set')

SHA256

4966C89216C02639A14EE3DA95580860E61EC223F0BFB7C1C03A142AC5986180

ModelTrends.ai Model ID

#6832

Sampler

DPM++ 2M Karras

Steps

20

Guidance

5

Seed

1437530670

Size

424x760

Clip skip

2

Sampler

DPM++ 2M Karras

Steps

20

Guidance

5

Seed

1437530663

Size

424x760

Clip skip

2

Sampler

DPM++ 2M Karras

Steps

20

Guidance

7

Seed

1511799189

Size

960x1920

Sampler

DPM++ 2M Karras

Steps

20

Guidance

8

Seed

3224160573

Size

384x768

Sampler

DPM++ 2M Karras

Steps

20

Guidance

7

Seed

1946587358

Size

960x1920

Sampler

DPM++ 2M Karras

Steps

30

Guidance

7

Seed

2250114057

Size

512x512

Clip skip

2

Sampler

DPM++ 2M Karras

Steps

30

Guidance

7

Seed

2661027254

Size

512x512

Clip skip

2

Sampler

Euler a

Steps

10

Size

3840x2160